For years, horseplayers who questioned computer assisted wagering have been given essentially the same warning. Be careful what you wish for. CAW represents too much money. Restrict it and handle will fall. Handle falls, revenue falls. Revenue falls, purses suffer. The tracks suffer. Horsemen suffer. Everybody suffers. I came to believe it myself. Today, I am not that sure.
NYRA finally decided to test part of that proposition, and the results from Saratoga are encouraging. They are also more complicated than the headline suggests.
Saratoga finished its 45 day summer meet with $905,425,541 in all sources handle, a record for the summer meet and a 3.5 percent increase over the comparable 2025 period. NYRA says CAW play declined approximately 29 percent while retail wagering increased. At the midpoint of the meet, the numbers were even more specific. Retail wagering had risen 6.8 percent, from $382.6 million to $408.7 million, while CAW wagering had fallen from $96.3 million to $68.4 million.
On its face, that’s exactly what many traditional horseplayers have been asking to see. Put some guardrails around high speed wagering, give the retail player a better experience, reduce the massive late price swings and see whether the regular customer responds. The regular customer appears to have responded.
Before we take a victory lap, though, there is a question that needs answering.
NYRA defines CAW activity according to the speed at which wagers are executed. Under its published policy, wagering is considered CAW activity when execution exceeds six bets per second. Inside the restricted period, CAW players can continue betting, but no faster than six wagers per second.
That isn’t theoretical. NYRA officials acknowledged earlier this year that some CAW groups stop betting after the cutoff while “quite a number” continue wagering at the permitted rate of six bets per second. NYRA also indicated that roughly 20 CAW players had been participating in its pools before the restrictions, with a significant portion of the subsequent decline attributable to two players leaving the pools altogether.
That raises an obvious accounting question. When an identified CAW player continues wagering at six bets per second or fewer, does NYRA count those dollars as CAW handle or retail handle?
I don’t know the answer, and that’s important. We aren’t going to accuse NYRA of reclassifying anything without evidence, and we shouldn’t assume that it does. NYRA may track identified CAW accounts separately from the technical definition it uses to restrict high speed activity. If that’s the case, fine. Say so, explain the methodology and the question disappears.
If, however, the classification of the reported handle changes when the speed of execution drops below the six wager threshold, that’s a very different story. Some portion of what we’re celebrating as increased retail wagering could actually be wagering generated by the same sophisticated customers operating under the new speed limit.
There is another wrinkle that no accounting definition can completely solve. CAW players aren’t computers. They’re horseplayers using computers. A sophisticated player can love a horse. He can love a sequence. He can see an exacta he thinks is badly mispriced or a Pick 5 he believes offers an extraordinary opportunity. If his expected advantage is large enough, the fact that he can’t make the wager through his normal high speed CAW channel doesn’t necessarily mean the wager disappears. He may be perfectly willing to surrender the rebate to get down on an opinion he believes is worth considerably more.
That wager could conceivably come through an ordinary account. It could come through another wagering channel. It could come through someone else’s account. History has already taught us that we would be foolish to assume every wager necessarily originates with the person whose name appears on the account.
We don’t even have to reach very far back into history for that lesson. The Marshall Gramm matter provided an uncomfortable reminder that account ownership, wagering direction and the ultimate source of sophisticated betting activity are not necessarily the same thing. That doesn’t mean anything improper is happening at Saratoga now, and it would be irresponsible to suggest otherwise. It means only that horse racing has already been shown why blindly equating an account classification with the person or operation behind the wagering can be dangerous.
A horseplayer would traditionally call one version of that betting through a beard. There can obviously be legitimate reasons for wagers to move among different accounts or channels as well. The larger point is the same. If money enters the pools through an account classified as retail, what exactly have we established? We know how the wager entered the system. We don’t necessarily know who generated the handicapping opinion or directed the money behind it.
That becomes particularly important when we’re being asked to draw conclusions from an apparent migration between CAW and retail handle. If some sophisticated money merely changes lanes, it hasn’t disappeared. It has changed labels.
None of this diminishes what may be the most important result of NYRA’s experiment. Late price volatility declined dramatically.
NYRA and the Thoroughbred Racing Protective Bureau developed a metric for measuring exacta pool volatility. Before the expanded guardrails, that index averaged 5.95. Afterward, it fell to 2.59, a reduction of more than 50 percent. That is real, measurable improvement for the wagering customer.
The guy who bets a horse at 5 to 2 and watches him win at 8 to 5 doesn’t care whether somebody calls the late money CAW, retail, institutional, sophisticated or Fred. He cares that the price he saw when he made his decision bears some reasonable resemblance to the price he receives when the race is over. For years, racing asked that customer to tolerate a wagering experience that would be unacceptable in almost any other financial transaction.
NYRA improved that experience, and they deserve credit for it.
NYRA also took a financial risk. CAW had represented approximately 20 to 22 percent of NYRA wagering before the expanded restrictions. After the changes, that share fell considerably. NYRA knew restricting some of its highest volume customers could cost money and proceeded anyway because it believed creating a more stable wagering environment was better for the long term health of the game.
Then Saratoga came.
At the halfway point, retail handle was reportedly up 6.8 percent while CAW handle was down approximately 29 percent. By the end of the meet, total handle had reached a record $905.4 million despite the substantial decline NYRA reported in CAW wagering. Those are encouraging numbers. They deserve attention throughout the industry.
They aren’t yet the end of the story.
Weather matters. Field size matters. Turf racing matters. Race quality matters. Saratoga being Saratoga matters. All of those things have to be considered before anybody assigns causation to the CAW restrictions. Nobody can responsibly say the guardrails caused Saratoga’s record handle simply because the two things occurred together.
The opposite is equally true. For years, one of the arguments against meaningful CAW restrictions was that reducing or inconveniencing this enormous source of wagering would necessarily hurt overall handle. Saratoga just provided a significant real world example in which CAW wagering reportedly fell substantially and overall handle didn’t collapse. It set a record.
That matters too.
Now we need to understand exactly what the underlying numbers mean, and the questions aren’t complicated. When NYRA reports that CAW wagering declined approximately 29 percent and retail wagering increased, how are wagers from identified CAW customers that continue wagering at six bets per second or fewer classified? Is the classification based upon the customer, the account, the wagering technology, the execution speed, the rebate arrangement or the individual transaction?
Those aren’t gotcha questions. They are the questions necessary to understand the experiment.
There is also a broader lesson here. For too long, the CAW debate has been reduced to two extremes. One side says computer players are destroying horse racing. The other says they are indispensable because they provide enormous handle and liquidity. Neither position gets us very far.
Sophisticated players aren’t the enemy. If someone builds a better model than I can, processes information faster than I can, handicaps better than I can and finds value that I missed, God bless him. Beat me because you’re better than me. That’s gambling. That’s handicapping. That’s competition.
What I object to is a structural advantage that the other customer cannot realistically overcome, particularly when that advantage affects the price after the other customer has already committed his money. There is a difference between being smarter than your opponent and being allowed to play by different rules.
The pari mutuel system doesn’t need the horseplayer to lose. It needs him to keep playing. Every dollar remaining in a player’s bankroll has the opportunity to become handle again and again. That’s churn, and it is one of the most important concepts racing routinely manages to forget.
A customer who believes the game is fair has a reason to come back tomorrow. A customer who believes he doesn’t know what price he’s getting when he presses the button eventually finds something else to do with his money. Racing spent years obsessing over maximizing today’s pool while perhaps not spending enough time thinking about maximizing the lifetime of tomorrow’s customer.
That’s why NYRA deserves credit regardless of where the classification question ultimately leads. They didn’t conduct another seminar about late odds movement. They didn’t commission another study or form another committee. They changed the rules, accepted the risk and produced a measurable reduction in late price volatility. That is real industry leadership, even at the potential risk of their self interests. That is seeing the big picture.
Now finish the job with transparency.
Tell horseplayers exactly how the CAW and retail buckets are constructed. Tell us how known CAW customers wagering below the six bets per second threshold are classified. Tell us whether the designation follows the account, the customer or the transaction. Give us enough information to determine whether retail wagering genuinely grew by the amount being reported or whether some sophisticated wagering may simply have migrated into another statistical category.
And understand that even perfect accounting cannot identify every sophisticated dollar. History has already shown us that. Rules can change. Accounts can change. Channels can change. People determined to get money into a wagering pool can be remarkably resourceful when there is enough money at stake.
History repeats. Horseplayers should know that better than anybody.
None of that changes the most encouraging part of Saratoga’s numbers. CAW handle, however NYRA ultimately defines it, reportedly declined substantially. Total handle set a record. Most importantly for the person actually trying to handicap and bet the races, late price volatility fell dramatically.
For years horseplayers were told what would happen if a major racetrack finally put meaningful guardrails around computer assisted wagering. NYRA did it, and the sky didn’t fall.
That’s progress.
Now show us exactly what we’re counting.
